The prognostic significance of intratumoral natural killer cells in patients with colorectal carcinoma

Natural killer (NK) cells have a spontaneous cytotoxic capacity against tumor cells. These cells represent a small proportion of human colon carcinoma‐infiltrating lymphocytes. Their prognostic significance in these tumors has yet to be determined.

rected to the inflammatory infiltrates as a reaction re-(NK) cells, can play an important role in the prognosis of intestinal tumors. 14,15 These cells have a particular flecting host resistance against malignancy. [5][6][7][8] Furthermore, lymphocytic infiltration has been consid-role in immunosurveillance against neoplasms and represent a small proportion of human colon carci-ered to be a good prognostic factor in colorectal carcinomas [9][10][11][12][13] and some authors have suggested that a noma-infiltrating lymphocytes. 16 The purpose of the current study was to evaluate subset of these lymphocytes with a spontaneous cytotoxic capacity against tumor cells, the natural killer the prognostic potential of intratumoral NK cell infil-

Other Histologic Variables Studied Patients
Pattern of growth One hundred eighty-six patients, (110 males and 76 Tumors were defined as expanding or infiltrating folfemales) aged 29 to 84 years (mean age: 63.4 years) lowing the morphologic criteria suggested by Jass et who underwent ''curative'' surgery for large bowel adal. 11 Expanding tumors were those with a circumenocarcinomas between 1977 and 1990 were studied.
scribed or pushing border of growth whereas infiltrat-No patient received adjuvant therapy after surgery.
ing tumors invaded in a diffuse fashion with an ill-Patients with synchronous cancer, inflammatory defined margin. There were 78 expanding tumors and bowel disease, or any type of multiple polyposis syn-79 infiltrating tumors. drome were excluded. Patients who died within 30 days after surgery or from causes unrelated to colo-

Degree of differentiation
Adenocarcinoma, well-differentiated indicated tumors rectal carcinoma were also excluded from the study, leaving 175 patients.

Peritumoral lymphocytic infiltration
The authors followed the criteria used by Jass et al. 11

Extensive infiltration (62 cases) indicated the presence Overall Survival
of an inflammatory band at the advancing tumor mar-Of the 157 tumors staged by TNM classification, 44 gin in which lymphocytes were the most abundant were Stage I, 58 were Stage II, and 55 were Stage III. cellular component. Absence or extreme paucity of The influence of stage on survival is shown by Kaplanlymphocytes (11 cases) was recorded as little infiltra-Meier curves (Fig. 2a) and Table 1 shows the corretion. An intermediate pattern of infiltration was desponding data. As expected, prognosis was signififined as moderate infiltration (84 cases).
In the study of NK cell intratumoral infiltration, a Number of involved lymph nodes significant difference in survival was found between N1 indicated 1 -3 metastatic lymph node (31 cases).
groups of patients with little or moderate versus those N2 indicated ¢4 metastatic lymph node (24 cases).
with extensive NK cell infiltration (P Å 0.0032) ( Table  2) (Fig. 2b). However, there was no significant difference between patients with little versus those with Follow-Up Patients were followed for at least 5 years or until moderate NK cell infiltration (P Å 0.388) ( Table 2). The statistical differences were significant (P Å death. Both overall survival and relapse free survival were examined. The overall survival was measured 0.0011) when the authors compared the cumulative survival curves of patients with little or moderate versus those from the time of resection of the tumor until death. Eight patients (5%) were lost to follow-up, 1 at 24 with extensive NK cell infiltration (Table 3) (Fig. 2c). The authors also studied the influence of NK cell months, 1 at 38 months, and 6 between 41 and 50 months. At 5 years, 63 patients (40.1%) had died, all intratumoral infiltration on the survival of patients with TNM Stage I, II, and III disease. Survival rates are with recurrent disease. Seventy patients developed recurrence during follow-up: 48 distant metastases and summarized in Table 4 and shown in Figures 2d, 2e, and 2f. The survival difference between patients with 22 local or regional recurrences. little or moderate versus those with extensive NK cell Disease Free Survival Disease free survival in relation to TNM stage infiltration was not statistically significant in patients with TNM Stage I and II disease (P Å 0.24 and P Å The 5-year relapse free survival rate of patients with Stage I tumors (n Å 44) was 89%, was 64% for those 0.11, respectively) but was very significant in patients with TNM Stage III disease (P Å 0.0008).
Step added P value Close alternatives infiltration; no other variables entered into the equa-Other variables considered: differentiation, pattern of growth, age ú 65 years, mucinous pattern, and tion ( Table 6).

Multivariate Analysis
gender.

DISCUSSION
The main purpose of this investigation was to evaluate in colorectal carcinomas the prognostic significance with Stage III tumors (n Å 55). The difference in the disease free survival rate was statistically significant (P of NK cells, a type of cell that forms part of tumoral inflammatory infiltrates. 16,19 Many authors have stud-õ 0.001). (Fig. 3a).

Disease free survival in relation to NK cell infiltration
The 5-year relapse free survival rate of patients with ever, because the cytotoxic activity of NK cells is activated by contact with neoplastic cells, 20 in the current extensive NK cell infiltration (n Å 25) was 80%, was 52% for those with moderate NK cell infiltration (n Å study only NK cells located in lymphocytic infiltrates among and around the tumoral glands were examined. 71), and was 49% for those with little NK cell infiltration (n Å 61). The difference in the disease free survival The authors chose the monoclonal antibody IOT-10 (CD57) for identification of NK cells because they analysis. However, although the NK cell infiltration has had previously observed that it immunostained these previously been studied in histopathologic material of cells consistently and accurately in paraffin embedded colorectal carcinomas, 16,19 to the authors' knowledge specimens. 21,22 This antibody recognizes a carbohythis is the first time that it has been correlated with drate antigen present on a subset of peripheral blood survival of patients. mononuclear cells involved in NK activity. The antigen In multivariate analysis of patients with Stage III is also expressed in neuroendocrine cells. In the curtumors, the number of involved lymph nodes was the rent study, the majority of CD57 positive cells in the first variable entered in the model. After that, the P lymphocytic infiltrates were found around the tumoral value of lympocytic infiltration increased from 0.001 glands. However, isolated CD57 positive cells occato 0.008 in overall survival and from 0.006 to 0.018 sionally were located very close to tumoral cells or in disease free survival, appearing as a weak ''close within the epithelium of neoplastic glands, similar to alternative'' to NK cell infiltration. After NK enters in the usual disposition of neuroendocrine cells of the the second step, lymphocytic infiltration becomes normal colonic epithelium. In these cases, the immunonsignificant (P value increased from 0.008 to 0.08 nohistochemical stain with chromogranin antibody in overall survival and from 0.018 to 0.124 in disease was always negative both in these cells and in the free survival). Lymphocytic infiltration is a clear ''close tumor cells.
alternative'' to NK cell infiltration as prognostic factor. With regard to the influence of NK cell intratu- The reason for that is the possible relationship bemoral infiltration on overall survival and disease free tween these two variables. survival at 5 years, the differences between survival However when the direct relationship between curves of patients with little or moderate versus those lymphoid reaction and NK cell infiltration was anawith extensive NK cell infiltration were very significant lyzed, no correlation was found between NK cell numin the current study. Therefore, extensive NK cell infilber and the degree of lymphocytic infiltration. For extration is a manifestation of a good immunoresponse ample, of 25 patients with extensive NK cell infiltraof the host against the tumor and is associated with a tion, 10 (40%) were found to have moderate or little/ good prognosis.
none lymphocytic infiltration and 47 of 132 patients Nevertheless, the authors wanted to examine with little or moderate NK cell infiltration (35.6%) whether the survival influence of NK cell intratumoral showed marked lymphocytic infiltration. This lack of infiltration could also be related to the tumoral stage.
correlation of one to the other is surprising. Although They found that patients with TNM Stage III tumors the reason for it is unclear, it may be related to differand extensive NK cell infiltration had significantly longer ent immunostimulating factors acting on the different survival than those with the same tumoral stage and subset of lymphoid cells present in colorectal carcinolittle or moderate NK cell infiltration. However, in spite mas. of the high survival rate of patients with TNM Stage I or Another type of lymphoid infiltration, ''Crohn's-II tumors and extensive NK cell infiltration, prognosis like lymphoid reaction'' also has been described as a was not related to the degree of NK cell infiltration in prognostic indicator in colorectal carcinoma. [24][25][26] The these patients. The reason for this is probably that a 5authors found this type of lymphocytic infiltrates in year follow-up is insufficient, especially in patients with only five cases, but this number of patients was too a generally high rate of 5-year survival, and it would be small to establish any correlation with NK cell infiltranecessary to continue the follow-up for a longer period tion and to statistically analyze patient survival. of time in this group of patients. Therefore, in the authors' multivariate model, To assess the relative prognostic value of the NK TNM stage and the number of metastatic lymph nodes cell tumoral infiltation, the authors studied other progwere the main prognostic variables in colorectal carcinostic variables in a multivariate analysis using the noma. However, the NK cell intratumoral infiltration Cox stepwise regression model (TNM stage, lymphoand the peritumoral lymphocytic infiltration are also cytic infiltration, pattern of growth, degree of differenindependent prognosticators of survival complementiation, mucinous pattern, age ú 65 years, and gentary to tumor invasion and lymph node metastasis. der). Three significant factors affecting overall survival In conclusion, the authors believe that in coloand disease free survival were selected in a stepwise rectal carcinoma, an extensive intratumoral infiltrafashion in increasing order as follows: TNM stage, NK tion of NK cells (ú three NK cells/field) that correcell infiltration, and lymphocytic infiltration. TNM 23 sponds to 17 NK cells/mm 2 , is associated with a favorstage and peritumoral lymphocytic infiltraable tumoral outcome and that NK cell intratumoral tion 10,11,12,24,25 have been related to prognosis in colorectal carcinoma in some studies using multivariate infiltration can be used as a variable with prognostic